Grid methodology for identifying co-regulated genes and transcription factor binding sites

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Identifying coordinately regulated genes is essential for mapping transcriptional networks.
• Bioinformatic approaches are needed to analyze complex gene expression data.

Purpose of the Study:

• To compare the performance of Markov clustering and k-means algorithms for gene clustering.
• To develop a method for identifying DNA-binding sites using gene expression data from co-regulated genes.
• To extend this methodology for cross-species analysis within a grid computing framework.

Main Methods:

• Comparative analysis of Markov clustering and k-means algorithms using microarray data.
• Development of a regression model linking gene expression levels to nucleotide pattern matching scores.
• Application of the model to identify DNA-binding sites in noncoding DNA sequences of co-regulated genes.

Main Results:

• Markov clustering demonstrated competitive performance against k-means in gene clustering tasks.
• The regression model successfully identified potential DNA-binding sites upstream of co-regulated genes.
• The framework was extended to accommodate multi-species analysis, enhancing its applicability.

Conclusions:

• Clustering algorithms like Markov clustering are effective tools in bioinformatics for gene network analysis.
• Integrating gene expression data with sequence analysis provides a powerful method for discovering regulatory elements.
• The developed approach offers a scalable and extensible framework for comparative genomics and regulatory network elucidation.